Fractally deforested landscape: pattern and process in a tri-national Amazon frontier.

摘要

Forest clearings in the Amazon are expanding along roads and are enhanced by the associated expansion of human settlements. The purpose of this research is to analyze the spatial patterns associated with this development process using fractal geometry and to partition this development process into different levels by a model-based classification scheme that can be applied to regions globally. A critical region of tropical forest cover in the tri-national frontier in the center of the southwestern Amazon was used as the study area. We utilized box-counting fractal dimensions to describe the spatial patterns of deforestation at a pixel level from 1986 to 2010 in the study region. The evolving pattern of development, as indicated by density-sliced fractal dimension, provides a unique and informative view of a deforesting landscape. The cleared areas have become increasingly compact from 1986 to 2010, where the low fractal dimensions typically represent little to no forest clearings and higher fractal dimensions are associated with more highly developed areas. Such differences are summarized by a classification scheme derived from a mathematical model that partitions the continuous range of fractal dimensions into five possible classes ranging from no or minimal development to highly developed. Such graphical representations of these stages of deforestation in the study region with such spatially explicit pixel-level information enables us to provide multi-level, local, adaptive, and flexible information to forest conservation groups, land managers and related programs.